Fix URI parser (paths were incorrect)

mhekkel · Apr 8, 2024 · 37fcbb9 · 37fcbb9
1 parent ed9cf27
commit 37fcbb9
Show file tree

Hide file tree

Showing 4 changed files with 40 additions and 20 deletions.
diff --git a/.gitignore b/.gitignore
@@ -13,3 +13,4 @@ debug/
 release/
 docs/api
 docs/conf.py
+src/revision.hpp
diff --git a/include/zeep/http/uri.hpp b/include/zeep/http/uri.hpp
@@ -89,11 +89,19 @@ class uri
 	~uri() = default;
 
 	uri(const uri &u) = default;
-	uri(uri &&u) = default;
-	uri &operator=(const uri &u) = default;
-	uri &operator=(uri &&u) = default;
 
-	void swap(uri &u);
+	uri(uri &&u)
+	{
+		swap(*this, u);
+	}
+
+	uri &operator=(uri u) noexcept
+	{
+		swap(*this, u);
+		return *this;
+	}
+
+	friend void swap(uri &lhs, uri &rhs) noexcept;
 
 	// --------------------------------------------------------------------
 

diff --git a/lib-http/src/uri.cpp b/lib-http/src/uri.cpp
@@ -102,16 +102,16 @@ uri::uri(const std::string &url, const uri &base)
 	remove_dot_segments();
 }
 
-void uri::swap(uri &u)
+void swap(uri &lhs, uri &rhs) noexcept
 {
-	std::swap(m_scheme, u.m_scheme);
-	std::swap(m_userinfo, u.m_userinfo);
-	std::swap(m_host, u.m_host);
-	std::swap(m_port, u.m_port);
-	std::swap(m_path, u.m_path);
-	std::swap(m_query, u.m_query);
-	std::swap(m_fragment, u.m_fragment);
-	std::swap(m_absolutePath, u.m_absolutePath);
+	std::swap(lhs.m_scheme, rhs.m_scheme);
+	std::swap(lhs.m_userinfo, rhs.m_userinfo);
+	std::swap(lhs.m_host, rhs.m_host);
+	std::swap(lhs.m_port, rhs.m_port);
+	std::swap(lhs.m_path, rhs.m_path);
+	std::swap(lhs.m_query, rhs.m_query);
+	std::swap(lhs.m_fragment, rhs.m_fragment);
+	std::swap(lhs.m_absolutePath, rhs.m_absolutePath);
 }
 
 std::string uri::string() const
@@ -324,14 +324,17 @@ const char *uri::parse_hierpart(const char *cp)
 
 			cp = parse_authority(cp);
 
-			if (*cp == '/' and cp[1] == '/')
+			if (*cp == '/')
 				m_absolutePath = true;
 
 			while (*cp == '/')
 			{
 				++cp;
 				cp = parse_segment(cp);
 			}
+
+			if (m_path.empty() and not m_host.empty() and m_scheme == "file")
+				throw uri_parse_error();
 		}
 		else
 		{

diff --git a/lib-http/test/uri-test.cpp b/lib-http/test/uri-test.cpp
@@ -59,7 +59,7 @@ BOOST_AUTO_TEST_CASE(uri_2)
 
 	BOOST_CHECK_EQUAL(url.get_scheme(), "http");
 	BOOST_CHECK_EQUAL(url.get_host(), "[::1]");
-	BOOST_CHECK_EQUAL(url.get_path().string(), "segment/index.html");
+	BOOST_CHECK_EQUAL(url.get_path().string(), "/segment/index.html");
 	BOOST_CHECK_EQUAL(url.get_query(false), "query");
 	BOOST_CHECK_EQUAL(url.get_fragment(false), "frag");
 }
@@ -68,14 +68,14 @@ BOOST_AUTO_TEST_CASE(uri_3)
 {
 	zeep::http::uri url("http://www.example.com/~maarten");
 
-	BOOST_CHECK_EQUAL(url.get_path().string(), "~maarten");
+	BOOST_CHECK_EQUAL(url.get_path().string(), "/~maarten");
 }
 
 BOOST_AUTO_TEST_CASE(uri_4)
 {
 	zeep::http::uri url("http://www.example.com/%7Emaarten");
 
-	BOOST_CHECK_EQUAL(url.get_path().string(), "~maarten");
+	BOOST_CHECK_EQUAL(url.get_path().string(), "/~maarten");
 }
 
 BOOST_AUTO_TEST_CASE(uri_5)
@@ -97,8 +97,16 @@ BOOST_AUTO_TEST_CASE(uri_6a)
 BOOST_AUTO_TEST_CASE(uri_6b)
 {
 	zeep::http::uri uri("file://a/b");
-	BOOST_CHECK(not uri.is_absolute());
-	BOOST_TEST(uri.get_path().string() == "b");
+	BOOST_CHECK(uri.is_absolute());
+	BOOST_TEST(uri.get_host() == "a");
+	BOOST_TEST(uri.get_path().string() == "/b");
+}
+
+BOOST_AUTO_TEST_CASE(uri_6c)
+{
+	BOOST_CHECK_THROW(zeep::http::uri("file://a"), zeep::http::uri_parse_error);
+	BOOST_CHECK_THROW(zeep::http::uri("file://a?b"), zeep::http::uri_parse_error);
+	BOOST_CHECK_THROW(zeep::http::uri("file://a#c"), zeep::http::uri_parse_error);
 }
 
 BOOST_AUTO_TEST_CASE(normalize_1)
@@ -236,7 +244,7 @@ BOOST_AUTO_TEST_CASE(encoding_1)
 	// http://a/höken/Ðuh?¤
 	zeep::http::uri u("http://a/h%C3%B6ken/%C3%90uh?%C2%A4");
 
-	BOOST_TEST(zeep::http::decode_url(u.get_path().string()) == "höken/Ðuh");
+	BOOST_TEST(zeep::http::decode_url(u.get_path().string()) == "/höken/Ðuh");
 	BOOST_TEST(zeep::http::decode_url(u.get_query(false)) == "¤");
 	BOOST_TEST(u.get_query(true) == "¤");
 }